Hierarchical Parallel Programming Model and Parallelization and Optimization Techniques Based on SMP Cluster
Zhenghua Wang
Abstract
Zhenghua Wang
Abstract
This paper describes the hybrid parallel programming model MPI/OpenMP which is suitable for the hierarchical architecture of SMP cluster in detail.It also provides the inter-and intro-node hierarchical parallelism.Then combined with an available performance evaluation method,a number of effective parallelization and optimization techniques on the MPI,OpenMP and single-processor three levels are introduced respectively.Meanwhile,we emphasize that performance optimization on single-processor is an important issue which cannot be ignored for improving parallel program performance.
OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
This paper describes the hybrid parallel programming model MPI/OpenMP which is suitable for the hierarchical architecture of SMP cluster in detail.It also provides the inter-and intro-node hierarchical parallelism.Then combined with an available performance evaluation method,a number of effective parallelization and optimization techniques on the MPI,OpenMP and single-processor three levels are introduced respectively.Meanwhile,we emphasize that performance optimization on single-processor is an important issue which cannot be ignored for improving parallel program performance.
Key concepts: Computer science, Parallel computing, Parallelism (grammar), Parallel programming model, Cluster (spacecraft), Programming paradigm, Node (physics), Programming language